Article ID: 126076, created on Jul 3, 2015, last review on Jul 3, 2015

  • Applies to:
  • Virtuozzo

Symptoms

After server reinstallation and MDS role recreation, the new MDS does not come online, and the following errors appear in pstorage -c cluster top:

11-08-14 09:42:48.731 Bad cluster id, expected 00076d8f4b46a0a26d4268494ae0f527
11-08-14 09:42:48.731 Target cluster has different cluster id than specified in cache at /root/.pstorage/cluster/mds_cache

Cause

During the server re-installation, a new cluster with the same name was created, instead of joining already existing cluster. Thus the conflict displayed by top appears: despite having same name, cluster ID is different.

Resolution

  1. Get the MDS ID assigned to the wrongly added server:

    # pstorage -c cluster list-services
    
  2. Remove the newly created MDS:

    # pstorage -c cluster rm-mds $MDS_ID
    

    ($MDS_ID should be replaced with the actual ID, obtained on the first step)

  3. Authenticate the node in cluster (cluster password should be supplied):

    # pstorage -c cluster auth-node
    
  4. Create a new MDS:

    # pstorage -c cluster make-mds -r /path/to/mds -a <ip>
    

Search Words

recreate metadata and chunk server

Target cluster has different cluster id than specified in cache

Bad cluster id

0dd5b9380c7d4884d77587f3eb0fa8ef 2897d76d56d2010f4e3a28f864d69223

Email subscription for changes to this article
Save as PDF